The study area was located in the southeastern fringe of the Tengger Desert (37∘28′ N, 105∘00′ E; elevation 1339 m). It lies within the transitional zone from desert steppe to steppified desert and also represents a transitional belt between desert and oasis. Based on meteorological records from 1956 to 2003, the mean annual temperature is 10.6 ∘C, with the minimum temperature being -25.1 ∘C in January and the maximum being 38.1 ∘C in July. The mean annual pan potential evaporation is around 3000 mm, while the mean annual precipitation is 180.2 mm, more than half of which falls in summer (June–August). The other three seasons typically experience more drought periods. The landscape of the study region consisted of large and dense reticulate barchan chains of sand dunes, where the predominant native plants were Hedysarum scoparium Fisch. and Agriophyllum squarrosum Moq. that together covered about 1 % of the ground surface. No biocrust was found on the surface of the mobile sand dunes (Li, 2012).

A no-irrigation vegetation system was established in 1956 to protect the Baotou–Lanzhou railway line from sand burial. It consisted of straw checkerboards as sand barriers to fix shifting dunes, with the subsequent planting of xerophytic shrub seedlings (Caragana korshinskii, Artemisia ordosica, Calligonum arborescens, etc.). The system was further expanded in 1964, 1981, and 1987. These vegetated areas were distributed parallel to the railway line, with a length of 16 km and a width of 1–2 km (Li et al., 2004). The initial shrub vegetation was gradually replaced by communities dominated by herbaceous plants due to the decreasing soil water content in the upper soil layers (Li et al., 2004). Biocrust biota then colonized and developed on the stabilized dunes, which resulted in the surface becoming increasingly stabilized. As a pioneer species, B. argenteum successfully colonized the revegetated area and became widespread, with a coverage exceeding 70 % on windward slopes and low-lying sand dunes. Although there is a gradual reduction in sand burial stress on the growth of B. argenteum as the biocrust moss becomes established, it is inevitably exposed to repeated wind-blown sand events, leading to dust burial of various depths. This burial is typically caused by two different processes that are seasonal in their severity. In spring, when the wind speed is usually the highest and drought is most severe (precipitation is the lowest), burial by wind-blown sand predominates. In summer and autumn, when the drought is slightly alleviated by higher levels of precipitation, animal activity (burrowing by ants, lizards, and rabbits) becomes important.

Samples of intact moss crusts (85 cm2, 10 cm thick) with 100 % coverage of B. argenteum were randomly collected using cylindrical PVC dishes (104 mm diameter, 12 cm height). At the base of each dish there was a drainage outlet that was covered with strips of nylon mesh to allow excess water to be removed, while preventing the loss of sand. All samples were collected from the interspaces between shrubs in the revegetated area that was established in 1981 and transferred to the Water Balance Observation Site (about 1 km from the sampling site) at Shapotou Desert Research and Experiment Station, Chinese Academy of Sciences. Sampling was conducted in late February and late August, which was about 10 days before the experiments began in spring and autumn in 2013, respectively. Samples were gently processed and sprayed with distilled water to ensure that they were moist and that the sample structure remained intact. All samples were placed below the ground surface, with the top 2 cm left aboveground. Rain shelters were then placed at a height of 2 m above the samples. The soil surfaces surrounding the samples were paved with a straw curtain, which extended for 5 m beyond the shade of the shelters to prevent disturbance from sand particles outside the study area.

Both drought and sand burial stress treatments were conducted from 10 March (spring) and 1 September (autumn). A total of 108 samples were collected for each experiment in the two different seasons and were randomly divided into three water supply groups by applying distilled water in three artificial rainfall regimes at 8-day intervals in spring and autumn: 4 and 6 mm (average rainfall, control), 2 and 3 mm (double drought), and 1 and 1.5 mm (4-fold drought), respectively. To determine the effect of burial, six treatments were applied, with depths of 0 (control), 0.5, 1, 2, 4, and 10 mm, equivalent to 0, 4.25, 8.5, 17, 34, and 85 mL of dried sand, respectively. The sand was distributed gently and evenly over crusts that had been subjected to each of the water supply subgroups described above.

There were six replicates of the drought×sand burial treatment. The prescribed sand burial depths and quantities of water applied were selected based on actual sand burial depths and precipitation levels observed during the period of 1990–2010 (Li et al., 2012) in the study area. The duration of each experiment was 72 days.

On the day after each experiment was completed, the sand particles deposited over the moss were gently blown off, and the same weight of sand applied prior to the burial treatment was collected. The upper 26 mm (i.e., including the active moss rhizoids) and inner core (5 cm diameter) were excavated from each original sample and placed into cylindrical plastic dishes (5 cm diameter, 28 mm height). Each dish had a drainage outlet at the bottom that was covered with a strip of nylon mesh to allow excess moisture to be removed. These smaller samples were more representative than the original samples because the edge effect of the PVC tube was removed.

The six small samples from each treatment were randomly divided into two subgroups: one for the determination of maximal shoot upgrowth, PSII photochemical efficiency, and regeneration potential and the other for the measurement of the chlorophyll a content of B. argenteum. The methods used to measure the maximal shoot upgrowth, PSII photochemical efficiency, regeneration potential, and chlorophyll a content of samples were adopted from Jia et al. (2012).

The maximal shoot elongation of B. argenteum was determined by the difference between the vertical distances from the upper edge of the PVC container to the uppermost part of the crust surface prior to sand burial and after removal from the sand at the end of the experiment using a Vernier caliper.

The samples were watered to saturation level and then cultured in a growth chamber (Thermoline Scientific Equipment Pty. Ltd, NSW, Australia). The photon flux density (PFD), air temperature (Ta), relative air humidity (RH), and CO2 concentration (Ca) were set to 1000 mmolm-2s-1, 25 ∘C, 55 %, and 390 mmolm-2s-1, respectively, during the day (08:00–19:00 UTC+8), and 0 mmolm-2s-1, 15 ∘C, 65 %, and 400 mmolm-2s-1, respectively, at night (19:00–08:00 UTC+8 the next day). The position of each sample was randomly changed every day. After a 3-day pre-acclimation, the samples were wetted again to saturation level and PSII photochemical efficiency was measured 4 h later (when the maximum value occurred). The PSII photochemical efficiency (Fv/Fm) was determined by an analysis of the slow kinetics of chlorophyll fluorescence using a PAM-2000 fluorometer (Walz, Effeltrich, Germany). The device was adjusted to maintain a distance of 1.20 cm between the fiber optics exit plane and sample. Prior to measurement, the samples were dark adapted for 5 min and then supplemented with a sequence of weak irradiance and saturation pulses (5000 mmolm-2s-1).

These samples were then hydrated, washed, and shaken to remove any sand attached. Then, 1 mm long portions of the upper shoots containing stem apices were isolated from each of the smaller samples and placed on native sand that had previously been sieved and dried as described by Stark et al. (2004). The same cylindrical plastic dishes as in the previous experiment were used, and the experimental conditions were also the same, except that moisture was supplied daily. The protonemal area was determined according to the protocols described by Barker et al. (2005) after a 58-day inoculation. The measurement of the regeneration potential of detached shoots was only conducted in the spring experiment, due to some of the autumn experiment samples being broken during transportation.

The chlorophyll a content was determined on a mgg-1 dry-weight basis by high-performance liquid chromatography (HPLC) using a method described by Gilmore and Yamamoto (1991). Briefly, 50 mg of dry, soil-free shoots were collected from each sample, ground, extracted in 80 % acetone, and then centrifuged at 10 000 rpm for 5 min. After the removal of the supernatant, the remaining pellet was resuspended in 100 % acetone and centrifuged again at 10 000 rpm. The supernatants were then mixed and passed through a 20 mm filter prior to injection into a Spherisorb ODS 1 column (Alltech Associates Inc., Deerfield, IL, USA) at a flow rate of 1 cm.

Three-dimensional plots were produced using MatlabR2014a (The MathWorks Inc., Natick, MA, USA) to show the responses of chlorophyll a content, PSII photochemical efficiency, regeneration potential, and shoot upgrowth of the biocrust moss B. argenteum to a combination of three levels of drought severity and six depths of sand burial.

A one-way analysis of variance (ANOVA) was used to test for any significant differences in the data using the SPSS 21 software (SPSS Inc., Chicago, USA). Differences between the individual parameters were evaluated by least significant difference (LSD) post hoc multiple comparisons at the 95 % confidence level.

A detrended correspondence analysis (DCA) of the chlorophyll a content, PSII photochemical efficiency, regeneration potential (protonemal area), and shoot upgrowth of B. argenteum was used to determine whether linear or unimodal ordination methods should be applied. We then performed a redundancy analysis (RDA) to determine the relationships between the parameters listed above and environmental parameters. A Monte Carlo permutation test (n=499) was used to determine the significance of all canonical axes. Both DCA and RDA were performed using Canoco for windows 5.0 (Ithaca, NY, USA).

Bryum argenteum can completely cover a soil surface, and there was no significant difference between spring and autumn cover when sand burial was absent (Table 1). Sand burial significantly reduced the B. argenteum cover in both seasons, with the amplitude decrease in autumn being significantly lower than that in spring (Table 1).

The chlorophyll a content of B. argenteum was generally lower in spring than under the same treatment in autumn, with the same trend found in the response to drought, sand burial, and their combination (Fig. 2). Drought uniformly imposed negative effects on the chlorophyll a content, whereas burial by sand had a dual effect on the chlorophyll a content (Fig. 2). The chlorophyll a content increased in treatments when the burial depth was shallow (<2 mm) and decreased when the depth was larger (sand burial depth ≥2 mm).

A significant interactive effect between drought and sand burial on the chlorophyll a content of B. argenteum was found. Drought strengthened the positive effects of shallow burial and mediated the negative effects of deep burial with regard to chlorophyll a retention (Fig. 2). In addition, sand burial weakened and even reversed the negative effects of drought on the retention of the chlorophyll a content in B. argenteum (Fig. 2).

The PSII photochemical efficiency of B. argenteum displayed the same trends as the chlorophyll a content in response to drought, sand burial, and their combination, although it was generally lower in spring than under the same treatment in autumn (Fig. 3). Drought consistently exerted negative effects on the PSII photochemical efficiency, while burial by sand had a dual effect on the PSII photochemical efficiency (Fig. 3). The PSII photochemical efficiency increased in treatments where the burial depth was shallow (<2 mm) and decreased when the depth was larger (sand burial depth ≥2 mm).

A dramatic interactive effect between sand burial and drought on the PSII photochemical efficiency of B. argenteum was observed. Drought strengthened the positive effects of shallow burial and ameliorated the negative effects of deep burial with regard to PSII photochemical efficiency (Fig. 3). Sand burial diminished and even reversed the negative effects of drought on the retention of the PSII photochemical efficiency (Fig. 3).

Drought imposed negative effects on the regeneration potential of detached shoots of B. argenteum, while burial by sand had a dual effect on the regeneration potential (Fig. 4). The regeneration potential increased in treatments where the burial depth was shallow (<2 mm) and decreased when the depth was larger (sand burial depth ≥2 mm).

There was a remarkable interactive effect of sand burial and drought on the regeneration potential of B. argenteum. Sand burial alleviated and even converted the negative effects of drought into positive effects with regard to the regeneration potential of detached shoots (Fig. 4). Drought enhanced the positive effects of shallow burial and eased the negative effects of deep burial on the regeneration potential of detached shoots (Fig. 4).

Although B. argenteum shoots were generally less elongated in spring than under the same treatment in autumn, drought and sand burial had both negative and dual effects on shoot elongation, which was similar to the pattern observed for the other three parameters described above (Fig. 5). Conversely, drought reduced the positive effects of shallow burial and exacerbated the negative effects of deep burial on shoot upgrowth. In addition, sand burial aggravated the negative effects of drought on shoot elongation (Fig. 5).

The RDA analysis results showed that drought was a more important stressor influencing shoot elongation than sand burial (Fig. 6), while sand burial played a more important role in the retention of the chlorophyll a content, PSII photochemical efficiency, and regeneration potential than drought. Specifically, sand burial was positively correlated with the chlorophyll a content, PSII photochemical efficiency, regeneration potential, and shoot elongation when the burial depth was shallow (<2 mm; Fig. 6a), while it was negatively correlated with the four variables when the depth was larger (sand burial depth ≥2 mm; Fig. 6b). In addition, drought was negatively correlated with the four variables when the burial depth was shallow (<2 mm; Fig. 6a) but positively correlated with all variables, except for shoot elongation, when the depth was larger (sand burial depth ≥2 mm; Fig. 6b).

The four parameters investigated in this study were more readily affected by sand burial and drought in autumn than in spring when the burial depth was shallow (<2 mm; Fig. 6a). Under deep-sand burial (sand burial depth ≥2 mm; Fig. 6b), the chlorophyll a content, PSII photochemical efficiency, and regeneration potential were more sensitive to sand burial in autumn than in spring, but shoot elongation was more susceptible to sand burial in spring than autumn. In contrast, the chlorophyll a content, PSII photochemical efficiency, and regeneration potential were more sensitive to drought in spring than in autumn, while shoot elongation was more susceptible to sand burial in autumn than spring under the deep-sand burial treatments (sand burial depth ≥2 mm; Fig. 6b).

As hypothesized, it was found that a combination of sand burial and drought did not always exacerbate the individual negative effects of each stressor on B. argenteum. Drought significantly ameliorated the negative effects of deep-sand burial on PSII photochemical efficiency (Fig. 2), the retention of chlorophyll a content (Fig. 3), and regeneration potential (Fig. 4) of B. argenteum. Sand burial diminished and even reversed the negative effects of drought on the maintenance of the chlorophyll a content (Fig. 2), PSII photochemical efficiency (Fig. 3), and regeneration potential (Fig. 4) of B. argenteum. These mutually antagonistic effects on the physiological vigor of the biocrust moss provided an opportunity for it to overcome the two co-occurring stressors, and this may be an important reason why it usually acts as the pioneer moss species by colonizing and even flourishing on the ground surface throughout China's sandy deserts.

The antagonistic effects of these two stressors are short-term physiological indicators, implying that B. argenteum has a strong potential to photosynthesize or regenerate after their removal. It is difficult for the moss to maintain this potential for a long time due to the increased use or exhaustion of its stored carbohydrate reserves when buried (Maun, 1998; Kent et al., 2005). Therefore, other long-term parameters (e.g., growth rate) also need to be considered.

In contrast to our expectations, the mutually antagonistic effects of drought and sand burial did not impact on long-term shoot upgrowth, even though sand burial (depth ≤4 mm) stimulated shoot elongation (Fig. 5). This additive negative effect inflicted by the combination of drought and sand burial on shoot upgrowth suggested a trade-off between growth and stress tolerance (Steinberg, 2012). In general, there is a trade-off between growth and physiological vigor, including regeneration potential, when the moss is exposed to stress. This is in accordance with the theory that adaptation to stress carries a cost, and spending resources on defense or resistance leads to a weakened performance in conditions where these traits are not needed (Bijlsma and Loeschcke, 2005). Collectively, the preservation of physiological activity (photosynthetic pigment, PSII photochemical efficiency) and the propagation of fecundity (protonemal area), afforded by the mutually antagonistic effects, at the cost of shoot elongation caused by the negative additive effect, under long-term, deep-sand burial, will result in the failure of B. argenteum shoots to protrude above the sand surface. This could even lead to death, ultimately causing the moss to vanish from the ecosystem. This can explain the absence of B. argenteum in areas suffering from long-term, deep-sand burial stress, such as flowing sand or seriously degraded landscapes. It also explains why B. argenteum can colonize soil surfaces only after the burial depth decreases to a shallow level, through ecological construction and restoration measures throughout the arid sandy areas of northern China.

Based on a conceptual model, Bowker et al. (2006) and Li et al. (2010) both proved that biocrust moss is restricted to a specific topography, where it is less stressed by the microclimate and disturbances than at other positions on micro-spatial scales. However, the distribution of B. argenteum is apparently vaster and more continuous than that indicated by Bowker et al. (2006) and Li et al. (2010) in arid sandy areas, where sand burial is pervasive and occurs regularly. Thus, it is suggested that the interaction between physical environmental stressors from resource limitation, climate, and physical disturbances can be used to facilitate an extension of the ecological niche of desert moss (Callaway, 1995). Therefore, corresponding models should take into account the interactions between climate change and physical disturbances, and from an evolutionary perspective the environmental pressure and biological response should be considered integratively. This study also found that biocrust moss could be harmed by climate change, with conditions such as drought predicted to be more frequent and extensive in the future. This damage may be alleviated by other environmental factors or disturbances, such as sand burial, although this needs to be verified further.

It is not fully understood why drought and sand burial exert antagonistic effects on the physiological activity and asexual propagating fecundity of B. argenteum. It is possible that the antagonistic effect may originate from the water-conserving effect of sand burial, which could mitigate the negative effect of drought to some extent (Meng et al., 2011). For over 300 years in China, sand burial has been widely used by farmers as a useful moisture-conserving measure to cultivate crops, in a practice referred to as Shatian or sandy field (Li et al., 2000). However, this beneficial effect has rarely been reported for biocrust mosses. In addition, sand burial can also provide a protective shell for B. argenteum, mitigating the damage from other stresses, such as wind (Liu et al., 2013). On the other hand, drought favors B. argenteum under sand burial by lowering the risk of carbon starvation induced by the reduction in photosynthetic area and the relatively trivial rainfall (causing the partial hydration of moss). This favorable effect of drought has also been reported for bundled filamentous cyanobacteria following sand burial (Williams, 2011; Rao et al., 2012) or thermal stress (Lan et al., 2014) and for mosses exposed to heat shock (Xu et al., 2009), ultraviolet-B (Turnbull et al., 2009), and fungal attack (Weber et al., 2016). Drought would increase the removal rate of sand by enabling the dry sand to be more easily blown by wind, resulting in a harmful deep burial becoming a beneficial shallow burial. Therefore, drought is also considered to have a dual effect, especially when deep-sand burial occurs.

The ability to achieve a higher rate of shoot elongation gives B. argenteum an important advantage over other moss species, enabling it to rapidly recover from sand burial (Jia et al., 2008). However, shoot elongation in B. argenteum is severely inhibited by drought, due to the lower amounts (Jia et al., 2008) and/or shorter durations (Kidron et al., 2010) of moisture availability. This could be interpreted as a reduction in the accumulation of carbohydrate gained by photosynthesis or as carbon starvation (Barker et al., 2005) caused by drought. Sand burial not only directly reduces the photosynthetic area of mosses but also causes a deterioration in the environmental conditions required for photosynthesis (e.g., the reduction in photosynthetically active radiation and blocking gas exchange). Furthermore, the sand deposited on B. argenteum would intercept water from precipitation, decreasing the quantity of rainfall available to the moss, with the trivial amount of precipitation received already identified as being detrimental (inducing carbon starvation) to biocrust moss (Alpert and Oechel, 1985; Belnap et al., 2004). This would consolidate the negative effect of drought on the shoot elongation of B. argenteum.

In recent years, the rapid artificial cultivation of biocrust has provided a novel alternative to traditional biological methods for controlling erosion (Bu et al., 2014; Doherty et al., 2015; Antoninka et al., 2016). At the same time, biocrust moss is considered to be a potentially promising biological material that could be inoculated to accelerate the process of sand fixation and the recovery of degraded soil (Antoninka et al., 2016). However, this technique is still limited to laboratory trials, with no successful large-scale application in the field reported. One key reason for this is that the moss typically occupies the late successional stage among biocrusts and its environmental requirements are high. The moss cultured in the laboratory under favorable conditions cannot withstand the unfavorable stress from drought, high temperature, and UV-B exposure. However, the results of this study indicate that moderate sand burial may have the potential to alleviate these stresses and increase the survival ratio of artificially cultured biocrust moss in the restoration of arid sandy deserts or degraded ecosystems. The use of such a technique is also in agreement with Maestre et al. (2006), who found that the inoculation of biocrusts in the form of slurry combined with the addition of composted sewage sludge, which has a similar effect to that of burial, encouraged the recovery of biocrust in degraded soils from semiarid Mediterranean areas.

Coverage (Table 1), physiological vigor (Figs. 2 and 3), and growth rate (Fig. 4) of B. argenteum and the response to drought and sand burial varied with season (Fig. 5). In our study area, the physiological activity of B. argenteum reached its lowest level after a long-term cold and drought stress in winter (Li et al., 2012). Conserving its activity in the continuous drought of spring enabled the moss to be ready to obtain more carbon through photosynthesis in the relatively favorable conditions in summer (higher precipitation). In autumn, when the physiological activity of B. argenteum was highest, it is essential to gain height through shoot upgrowth to cope with the following sand burial in winter. This successful seasonal adaptation strategy of B. argenteum to the co-occurring stressors of drought and sand burial was supported by our results. Therefore, the seasonal distribution of precipitation and sand burial in our study area was important because it enabled B. argenteum to be the pioneer species in our study area, and this mechanism may be valid in areas suffering from the co-occurring stressors of drought and sand burial in sandy deserts elsewhere in China and worldwide.